Organizational Learning from Performance Feedback: A Behavioral Perspective on Multiple Goals

3.1 The Conundrum of the Kinked-Curve Relation

Early formulations of the theory did not specify the functional form of the relationship between performance in relation to an aspiration level and the propensity to search and make changes. This left open the possibility that changes in performance over the entire range of performance would have the same effect on these organizational behaviors (e.g., Reference Lant, Milliken and BatraLant, Milliken, and Batra, 1992; Miller and Chen, 1994). Reference GreveGreve (1998), in an influential study, suggested that responses to performance in relation to an aspiration level differ depending on whether performance is above or below an aspiration level. Specifically, he proposed that the impetus for change stemming from unsatisfactory performance is fraught with more resistance than the commitment to the status quo arising from satisfactory performance. Unsatisfactory performance plants the seed for motivating the search for new solutions and change, but the extent to which these behaviors materialize hinges on the strength of other processes that underlie resistance to change. Greve found initial evidence of this discontinuity or kink in the slope of the relationship between performance in relation to an aspiration level and change in a study of radio stations. Audience share below the aspiration level increased the probability of the introduction of new radio formats less strongly than audience share above it decreased it, producing a kinked-curve relation from audience share to format change. Later empirical work found additional evidence of change behaviors being less sensitive to performance below an aspiration level than above it. A study of shipbuilders found that performance above an aspiration level decreased the propensity to launch innovations, whereas performance below an aspiration level did not have any effect (Reference GreveGreve, 2003a). Likewise, a study of hard disk drive producers found that the propensity to introduce new hard disk drives was negatively impacted by performance above an aspiration level but it was not significantly affected by performance below an aspiration level (Reference Audia and BrionAudia and Brion, 2007: Study 2). Works such as Baum and Dahlin’s study of railroad firms (2007) and Diwas, Staats, and Gino’s study of cardiac surgery (2013) lend additional support by reporting that both organizations and individuals learn more from high performance than from low performance.

The fact that organizations initiate searches for new solutions or make changes less readily in response to performance below an aspiration level than previously assumed poses a theoretical conundrum that current formulations of the theory resolve by suggesting that managers’ attention is not necessarily fixed at all times on an aspiration level. Attention shifts between different standards of evaluations and these shifts yield variations in responses to unsatisfactory performance. The theoretical question then becomes: What mechanisms can account for these shifts of attention?

3.2 Responsiveness to Performance and the Survival Point

The first account draws on Reference March and ShapiraMarch and Shapira’s (1992) shifting focus model of risk-taking, which introduces the notion of a survival point as an alternative standard of evaluation. March and Shapira propose conditions under which managers shift their standard of evaluation from an aspiration level to a survival point defined as the point at which all resources are depleted and the organization fails. Drawing on theories of threat rigidity (Reference Staw, Sandelands and DuttonStaw, Sandelands, and Dutton, 1981) and risk (Reference LopesLopes, 1987), they suggest that when such a shift occurs managers become risk-averse. The rationale is that attention to a survival point implies a heightened concern for security that deters managers from taking actions that have uncertain outcomes and thus may lead to catastrophic outcomes.

Reference March and ShapiraMarch and Shapira’s (1992) analysis is not strictly about how performance is assessed but rather about how managers assess an organization’s stock of resources. Indeed, its theoretical connection to the study of responsiveness to performance feedback was made more than ten years after its publication. Drawing on March and Shapira’s shifting focus model, Reference Audia and GreveAudia and Greve (2006) suggest that whether managers focus their attention on an aspiration level or on a survival point depends on which of the two is closer to actual performance. In their analysis, attention goes to the standard closest to performance because the consequences of reaching the closer point loom larger. So, when low performance is closer to the survival point, the concern for avoiding a catastrophic outcome outweighs the desire to improve performance. Low performance is seen as a step closer to failure rather than as a repairable gap. Fear and the need for safety prevail over the desire to improve (Reference LopesLopes, 1987). Audia and Greve view an organization’s stock of resources as a key influence on the position of a survival point, both cognitively and in reality, because resources such as financial assets, manufacturing infrastructures, and a large workforce allow organizations to endure many periods of poor financial performance with little threat of failure (Reference LevinthalLevinthal, 1991). This buffering effect of a large stock of resources lowers the performance level at which the organization’s survival is in danger, in essence rendering a survival point more distant and less salient. Small resource endowments, in contrast, raise the level of an organization’s survival point and thus heighten awareness of the distance between performance and the survival point.

Taking an organization’s size as a proxy for the stock of resources, they suggest that managers of small and large firms will respond differently to performance below an aspiration level. When compared to managers of large firms, small firms perceive that performance below an aspiration is closer to the survival point. These perceptions should prompt a greater concern for security and a diminished propensity to take actions with uncertain outcomes such as those involving the search for new solutions and the introduction of change. Given these heterogeneous responses, what may appear in aggregate reduced organizational responsiveness to low performance or insensitivity to low performance may mask variations in responsiveness of small and large organizations.

Reference Audia and GreveAudia and Greve’s (2006) empirical analyses of the factory expansion of shipbuilders support this argument. Large shipbuilders generally invest in factory expansion less than small shipbuilders, a finding consistent with the idea that bureaucratic constraints make large firms inert and risk-averse, but large shipbuilders increase factory expansion more than small shipbuilders in response to performance below an aspiration level. Reference GreveGreve (2011) replicates these results in a sample of shipping firms. In response to low performance, large shipping firms, compared to small firms, acquire more ships as well as ships more different from the ones they already have. Both studies also reveal decreased propensities to make changes when performance is above an aspiration level. Whereas these studies focus on size as a proxy of the stock resources, later empirical work extends this reasoning to other resources that may serve as buffers from threats to the survival of the organization. Reference DesaiDesai (2008) examines positive media coverage as a proxy for legitimacy in a study of capacity expansion of railroad firms, whereas Reference Lehman, Hahn, Ramanujam and AlgeLehman et al. (2011) focus on deadline proximity as an indicator of time scarcity in a study of fourth-down decisions made by football teams. In both instances, the findings are supportive of variations in responsiveness to low performance as a function of resources that buffer from catastrophic outcomes: A smaller stock of resources is associated with a more muted response to low performance, in line with empirical evidence of a kink in the slope of the relationship between performance in relation to an aspiration level and the propensity to make changes.

3.3 Self-enhancing Assessments of Low Performance

A second account of organizations’ lesser responsiveness to low performance comes from research that draws on the social psychological literature on self-enhancement, people’s desire to view the self, and to have others view the self, in the most positive light possible (Reference KundaKunda, 1990; Reference Sedikides and StrubeSedikides and Strube, 1997; Reference Pfeffer and FongPfeffer and Fong, 2005). A key contingency in this work is the ambiguity that is often encountered when assessing performance (Reference Audia and BrionAudia and Brion, 2007). Research on self-enhancement holds that feedback that is ambiguous, meaning that it is open to multiple interpretations, is often interpreted in favorable ways because favorable interpretations of performance enable individuals to maintain a positive image (Reference Dunning, Meyerowitz and HolzbergDunning et al., 1989). The inclination to tackle performance deficits by searching for new solutions and making changes is still important but it is especially relevant when performance is unambiguously unsatisfactory (Reference Jordan and AudiaJordan and Audia, 2012; Reference Audia, Brion and GreveAudia, Brion, and Greve, 2015). In unambiguous situations, self-serving assessments of low performance would be difficult to defend because evaluative assessments reflect not only what people want to believe but also what dispassionate observers consider credible (Reference KundaKunda, 1990).

Ambiguity surrounding performance assessment may arise in different ways. Much of the empirical research reviewed earlier focuses on a single performance measure but when other performance measures are taken into account ambiguity may arise as a result of performance measures that give different signals. Reference Audia and BrionAudia and Brion (2007) examine how individuals and organizations respond to situation in which a secondary performance measure diverges from a primary performance measure. Consideration of two performance measures that vary in importance is particularly revealing because, if the desire to be accurate guides behavior, a primary performance measure should hold more weight in guiding behavior than a secondary performance measure. This implies that, when performance on a primary performance measure is low, it should not make any difference whether a secondary performance measure is above or below an aspiration level. Audia and Brion find, however, that a primary performance measure that falls below an aspiration level instigates less change when a secondary performance measure is above an aspiration level than when a secondary performance measure is also below an aspiration level. In addition, they find that when a secondary performance measure is more favorable than a primary performance measure, individuals rearrange the hierarchy of the two performance measures giving more importance to the secondary performance measure. Reference Audia and BrionAudia and Brion (2007) find supporting evidence of self-enhancing interpretations of diverging performance measures and lower responsiveness to performance below an aspiration level both in an experimental study of individual decision-makers (Study 1) and in an archival study of disk drive producers (Study 2). Reference Blagoeva, Mom, Jansen and GeorgeBlagoeva, Mom, Jansen, and George (2020) report additional evidence of self-enhancing responses to performance on multiple goals in a study of R&D investments.

Ambiguity may also result from performance comparisons to multiple aspiration levels. Reference Joseph and GabaJoseph and Gaba (2015) study divergences in aspirations over time in the global mobile phone industry. They argue that managers take into account how market share changes over time in relation to both historical and social aspiration levels and respond differently to feedback that is consistent (positively correlated), inconsistent (negatively correlated), and ambiguous (weakly linked). Consistent feedback sends an unequivocal signal of either improvement or decline and should trigger decreases or increases in the propensity to make change, respectively, in line with the predictions of the theory. Inconsistent feedback is also easy to interpret, they propose, because it is difficult to paint rosy pictures of situations in which organizations are improving over their past performance but at a slower rate than that of their competitors or when they are doing better than their competitors but worse off than their past performance. In both instances, they predict that inconsistent feedback increases change. Ambiguous feedback, evidenced by performance patterns that are neither distinctly improving nor declining, creates greater opportunities for favorable interpretations. Under conditions of ambiguity, managers perceive that they can form defensible favorable assessments by, for example, giving greater weight to information that is positive and casting doubt on information that is negative. Because of these favorable interpretations, Joseph and Gaba suggest that ambiguous feedback should reduce responsiveness to performance. Their empirical analyses of global mobile phone firms’ propensity to introduce new products in response to their market share support these predictions. Reference Lucas, Knoben and MeeusLucas, Knoben, and Meeus (2018) find a similar pattern of self-enhancing responses to performance feedback when diverging aspirations introduce ambiguity in performance assessment. Reference Audia, Brion and GreveAudia, Brion, and Greve (2015) lend additional support to self-enhancing responses to low performance by showing in an experimental setting that, when performance is low, decision-makers select social comparisons that are favorable. Reference Smith and ChaeSmith and Chae (2017) extend their finding showing how self-enhancing selections of favorable social comparisons are enabled by the ambiguity of organizational forms.

3.4 Integrating Survival Point and Self-enhancement Accounts of the Kinked Curve

To summarize, two key observations emerge from this body of work. First, there is considerable evidence supporting the view that responses to performance in relation to an aspiration level differ depending on whether performance is above or below an aspiration level. Specifically, the impetus for change stemming from unsatisfactory performance is fraught with more resistance than the commitment to the status quo arising from satisfactory performance (Reference GreveGreve, 1998). Unsatisfactory performance plants the seed for motivating the search for new solutions and change but the extent to which these behaviors materialize hinges on the strength of other processes that underlie resistance to change. Second, two distinct mechanisms contribute to accounting for the kink in the relationship. The first mechanism concerns switches of attention away from an aspiration to a survival point that are likely to occur when performance is low, and organizations have a limited stock of resources. These shifts of attention are concomitant to a heightened concern for safety that deters managers from taking actions that have uncertain outcomes and thus can cause the organization to fail. The second mechanism concerns self-enhancing interpretations of low performance rendered possible by ambiguity arising from diverging performance measures or from weakly linked performance assessments stemming from multiple aspiration levels. In these instances, ambiguity enables managers, who are generally inclined to portray negative outcomes in a favorable way, to formulate defensible self-serving assessments.

The prevalent approach in recent empirical work has been to identify contextual features such as an organization’s stock of resources and sources of ambiguity in performance assessment that heighten the influence of, respectively, a survival point or self-serving assessments of performance. Research addressing conditions that activate these mechanisms has proceeded separately but there are interdependencies that require consideration. Both self-enhancement and attention to the survival point predict decreased responsiveness to low performance but the theory suggests different conditions triggering these two distinct mechanisms. Self-enhancement is enabled by ambiguity (Reference Audia and BrionAudia and Brion, 2007; Reference Joseph and GabaJoseph and Gaba, 2015). When ambiguity is low, self-enhancing interpretations are less likely and the problem-solving mode is likely to guide action. Under conditions of low ambiguity, when room for favorable interpretations of low performance is lacking, low performance may trigger a focus of attention on the survival point when decision-makers perceive a threat to vital interests (Reference Audia and GreveAudia and Greve, 2006). This is likely to happen when the stock of resources is depleted to a point that additional instances of unequivocal low performance create a precarious state for the organization and/or the members responsible for the performance. Under those conditions, resources that buffer from safety concerns are expected to impact attention to the survival point. The stock of tangible assets creates differences in perceptions of safety among small and large firms that are equally low performing. Similarly, the degree of support from external stakeholders accounts for variations in perceptions that safety is at risk when performance is low (Reference DesaiDesai, 2008) and sport teams feel closer to the survival point when time is running out during a match (Reference Lehman, Hahn, Ramanujam and AlgeLehman et al., 2011). So to infer whether low responsiveness to low performance stems from self-enhancing assessments or from attention to the survival point, the modified core model suggests considering the degree of ambiguity confronted by decision-makers who assess performance and the stock of resources indicative of proximity to the survival point.

Two types of studies are particularly valuable as researchers seek to identify when self-enhancement or attention to the survival point prevail in influencing responses to low performance. Longitudinal studies of organizations (e.g., Reference GreveGreve, 2008) and individuals (e.g., Reference Audia and GoncaloAudia and Goncalo, 2007) in which attention to different standards of evaluation is not directly examined but is inferred from the kind of responses to performance researchers observe in the data will continue to be a valuable source of insights. These studies can identify and test the influence of characteristics of the organizations and the environments that are theoretically linked to distinct mechanisms. The modified core model, with its greater emphasis on variations in the process of performance assessment, also calls for research that directly examines how organizations form standards of evaluation and how they direct their attention to them (e.g., Reference Audia, Brion and GreveAudia, Brion, and Greve, 2015; Reference Smith and ChaeSmith and Chae, 2017). This research can include examination of organizations with goals that are distinct because they are not conventional firms but rather hybrid organizations, voluntary organizations, or governmental organizations.

Research on goal and aspiration level formation is a direction that was pursued early on by researchers interested in performance feedback (e.g., Reference LantLant, 1992) and that has started to gain renewed interest. The research has addressed a wide range of questions. First, there are alternative models of aspiration level updating, all of which originate from Cyert and March but make some modifications such as entering an upward drift or separating the historical and social aspiration level. Because the first studies were experimental, later work added investigations of aspiration level updating using data from firms (Reference Mezias, Chen and MurphyMezias, Chen, and Murphy, 2002). Other work has tested different aspiration level specifications for fit against broad databases of firms (Reference Bromiley and HarrisBromiley and Harris, 2014). Still other work investigates aspiration levels that combine social awareness and historical performance in new ways (Reference Moliterno, Beck, Beckman and MeyerMoliterno et al., 2014), identifies specific effects of social and historical performance comparisons (Reference Kacperczyk, Beckman and MoliternoKacperczyk et al., 2015), or captures interorganizational differences in aspiration level updating (Reference Blettner, He, Hu and BettisBlettner et al., 2015). The findings have not fundamentally altered how aspiration levels are modeled, but they have pointed to ways in which empirical measurement and theory can be improved.

The modifications of the core model of performance feedback are important for at least two reasons. The first is that they provide theoretically driven and empirically supported accounts of the long-standing puzzle of why organizations often fail to learn from failure. Clearly, the extent to which a theory of performance feedback offers persuasive and testable explanations of organizations’ lack of responsiveness to low performance is a key indication of its strength. The second is that they integrate into the model of performance assessment originally developed by Cyert and March well developed concepts such as the survival point and self-enhancing interpretations of low performance that offer additional opportunities for expanding the predictive power of the model. An important by-product of the work that tackles the conundrum of the kinked curve is that it brings renewed attention to multiple goals. Indeed, according to the modified core model, muted responses to low performance are often due to shifts of attention across goals. These shifts of attention were an important insight in the original version of the theory. Next, we turn our attention to the component of the theory regarding the influence of multiple goals on search and change and discuss theoretical advances that expand on these original insights. We will find that research on multiple goals, although still in its infancy, has already made enough advances to suggest opportunities to create an extended model of performance feedback and provide guidance on what its components should be.

4.1 Vertical Differentiation of Goals

Organizations are hierarchical systems that coordinate behaviors at different levels, learn at different levels, and have goals at different levels. The structure of goals within organizations was an important topic in early work (Reference March and SimonMarch and Simon, 1958), with a focus on how high-level goals could be decomposed to sub-goals at lower levels that let boundedly rational decision-makers reach a nearly optimal set of behaviors from the organizational point of view. This differentiation of goals was less prominent in the behavioral theory of the firm, which had less of a focus on organizational design and more on organizational behavior (Reference Cyert and MarchCyert and March, 1963). Research is returning to this issue now that the observation of goals at different levels has spurred interest in how they interrelate.

Vertical differentiation of goals has also been a topic of recent research, but without the goal composition element of the early work. Instead, goals at different levels have been seen as potentially reinforcing and potentially rivaling. This research is built on observations such as the presence of the same goal (ROA) at both corporate and division levels (Reference Gaba and JosephGaba and Joseph, 2013), as well as profitability goals at higher levels interacting with other goals at lower levels (Reference Audia and SorensonAudia and Sorenson, 2001; Reference Joseph, Klingebiel and WilsonJoseph, Klingebiel, and Wilson, 2016). The findings suggest a clear tension between the relevance of a lower-level goal and the hierarchical (and survival-related) priority of higher-level profitability. At its current stage, however, this research is more an invitation to further theory development than a conclusive statement. Because the higher-level goal in the studies so far has been ROA, which is known for its great behavioral impact regardless of organizational level, it is unclear whether its effect strength reflects the high hierarchical position or the nature of the goal.

4.2 Horizontal Differentiation of Goals

Intraorganizational goals are also horizontally differentiated, and again this can be seen as a consequence of overall goals being decomposed. Indeed, the listing of sales, inventory, and production goals by Cyert and March (1963: 117) suggests exactly such a decomposition toward the main goal of profitability. Organizations deal with the dilemma that each goal is associated with one (or more) organizational units, however, and often have structures that do not encourage and facilitate coordination among the multiple decomposed goals. It still seems obvious that coordination occurs at some level, as the juxtaposition of these three goals suggests, but this leaves open the question of how the organization responds to falling behind on any of these goals, either as a stated aspiration level or one with a historical or social source. Research on horizontally differentiated goals is completely absent as far as we have been able to determine, as the focus has been on comparisons of vertically differentiated goals. This is probably because vertically differentiated goals are a more salient feature of the empirical reality of interest to management scholars, but the low interest in horizontally differentiated goals is still a striking omission.

4.3 Externally Imposed Goals

The single most important advance in organizational theory since Reference Cyert and MarchCyert and March’s (1963) behavioral theory of the firm has been the increased focus on the organizational environment as an influence on organizational behaviors (Reference ScottScott, 1987). Interestingly, work on how performance and aspiration levels affect organizational behaviors has been late to incorporate these insights and has instead followed the original focus on internal organizational goals. However, there is increased realization that organizations also face externally imposed goals, starting with the observation that institutions often impose rule-like compliance and monitoring by external actors (Reference Meyer and RowanMeyer and Rowan, 1977) and continuing to the recent attention to the effects of external rating and ranking systems on organizations (Reference Sauder and EspelandSauder and Espeland, 2009). Although empirical examination of the effects of external goals is limited, recognition of these effects has led external actors to use public rankings to change organizational behaviors (Reference Bartley and ChildBartley and Child, 2011). Initial evidence suggests that such interventions have some success (Reference Chatterji and ToffelChatterji and Toffel, 2010; Reference Sharkey and BromleySharkey and Bromley, 2015; Reference Rowley, Shipilov and GreveRowley, Shipilov, and Greve, 2017; Reference Shipilov, Greve and RowleyShipilov, Greve, and Rowley, 2019).

4.4 Behavioral Consequences of Multiple Goals

Reference Cyert and MarchCyert and March (1963) proposed two solutions to a scenario in which organizational members react to performance on multiple goals. The first is sequential attention to goals. Because they do not have the cognitive capacity to simultaneously attend to multiple goals, managers cope with their bounded rationality by tackling goals sequentially. They start by tackling the goal highest in a hierarchy of importance. When the performance on that goal is satisfactory, they move their attention and effort toward the pursuit of the next goal in the hierarchy. Underlying the idea of sequential attention to goals are the assumptions that a goal hierarchy exists and that managers display consistency in adhering to this goal hierarchy.

The second solution to the multiplicity of goals sheds light on how a goal hierarchy emerges. Reference Cyert and MarchCyert and March (1963) suggest that groups inside organizations have conflicting interests that are often reflected in the importance they assign to different goals. Through a process of negotiation and power accumulation, a dominant coalition emerges that temporarily shapes goal priorities. So, when more than one goal can be plausibly seen as most critical for the organization, the dominant coalition addresses the potential goal conflict by giving greater weight to the goal that aligns with the views of its members. That goal becomes the primary concern even if other groups within the organization may have different preferences. Importantly, Reference Cyert and MarchCyert and March (1963) highlight the unstable nature of this internal equilibrium, noting that it involves continuing negotiation and that it depends on side payments of resources in return for support from groups. In essence, goal conflict is often unresolved and inconsistency among goals complicates understanding of how organizations respond to multiple goals.

Current theory has moved beyond these core ideas regarding multiple goals. New insights have emerged at different levels of analyses. Because the majority of the empirical work on learning from performance feedback focused on responsiveness to performance on a single goal, multiple goals were initially brought in to address the puzzle of the kinked curve and more generally of muted responses to low performance. Later research examined multiple goals because they are such a well-known feature of organizational life, but we believe that their theoretical implications have not yet been developed fully. In this Element we propose that multiple goals should become more prominent in the study of responsiveness to performance feedback. Taking this perspective, instead of assuming the primacy of a single goal, researchers would start by considering multiple goals that are important to organizations and their subunits and seek to explain why and when decision-makers across the organizational hierarchy respond to performance on a particular goal. Such a theoretical and empirical move would help develop a deeper understanding of how features of the organizational context and the external environment alter attention and responsiveness to goals. In Section 5, we develop an extended model of performance feedback building on theory at the individual level of analysis, but also taking into consideration influences at the organizational and environmental levels of analysis. The purpose is to outline promising areas of research and theoretical mechanisms that need further exploration.

5.1 Individual Level

We begin by developing the individual-level component of such an extended model by identifying three key features that influence how decision-makers respond to multiple goals: personal involvement, ambiguity, and power. Our analysis suggests that self-enhancing responses to multiple goals are most likely when these three features have high values. Low values on any one of these three features put a brake on self-enhancing responses and make a problem-solving response more likely. After we specify this individual component of the model, we build on it to propose how the organizational hierarchy and features of the external environment impact the prevalence of a self-enhancement or problem-solving orientation among subunit managers and top managers.

5.2 Organizational Level

Starting with Reference Cyert and MarchCyert and March (1963), much of the organizational literature on performance feedback views the organizational hierarchy as consisting of two hierarchical levels: the organizational level and the subunit level. At the organizational level, top managersFootnote ² are responsible for achieving organizational goals and responding to performance on those goals by making changes to the strategy and administrative structure of the entire organization. Top managers also decompose organizational goals into goals assigned to subunits and develop control systems aimed at achieving coordination among subunits. Subunit managers have narrower job domains as they are expected to coordinate activities within subunits, pursue subunit goals, and adjust their behavior in response to performance on such goals. The administrative mechanisms used by top managers to coordinate and control organizational activities are an important influence on how subunit managers assess and respond to performance on multiple subunit goals.

Building on the individual component of the extended model we outlined earlier, we identify two features of the formal organization that alter subunit managers’ propensity to adopt a self-enhancement or problem-solving orientation: the centralization of subunit decisions and the formalization of subunit performance evaluation. Both limit self-enhancement and make a problem-solving response to performance on multiple goals more likely. Centralization of subunit decisions weakens the link between subunit low performance and self-threat to the people participating in decision-making, whereas formalization of subunit performance evaluation reduces ambiguity, thereby limiting decision-makers’ latitude to form interpretations of performance that are credible in the eyes of others. We also identify one feature of the informal organization that alters subunit managers’ choice of orientation: the power of their subunit. Regardless of the formal authority relations and processes for managing these, power is distributed unequally in organizations and affects subunit managers’ compliance with the formal authority (Reference Salancik and PfefferSalancik and Pfeffer, 1974; Reference PfefferPfeffer, 1981).

5.3 Environment Level

Continuing to build on the individual component of the model, we now turn to an analysis of how the environment influences an organization’s responsiveness to multiple goals. Our focus is on variations in environmental influences that alter the degree of ambiguity of environmental demands and in so doing enable or constrain top managers’ self-enhancing responses to multiple goals thereby making a problem-solving orientation more likely. We start by drawing on institutional research that views environments as potentially consisting of multiple logics. The concept of logics is important as we seek to develop a theoretical perspective on multiple goals because it ties cultural constructions that infuse organizational behaviors with meaning to the process by which organizations define and pursue goals. We continue by noting that the environment also differs in the formalization of performance assessment, which influences the managerial ability to engage in self-enhancement. Finally, organizational power relative to the environment has similar effects as subunit power relative to the top management, as it allows the organizational top management to disregard some of the influence that the environment gains from having clearly defined and formalized goals and external assessment.

5.4 Full Extended Model

Figure 2 illustrates the extended model in simplified form. As the figure shows, performance on multiple goals can result in a problem-solving response or a self-enhancing response, and individual, organizational, and environmental factors influence which type of response is more likely. Seeing these two types of responses as mutually exclusive at all times is a simplification since decision-makers in organizations are capable of engaging in problem-solving and self-enhancing actions at once. Indeed, even a manager who fully intends to engage in a self-enhancing reaction to performance feedback may undertake a problemistic search as a form of cover. After all, examining alternatives is the best way to argue that the current set of actions is the best choice, and that all other choices would have given even worse performance than the current. A subunit manager making such a presentation in a formalized and centralized organization may succeed and sometimes be able to engage in self-enhancement (contrary to our prediction), but may also fail and be forced to make changes, or may even be replaced by someone else.

Figure 2 Extended model of performance feedback

As the figure shows, other elements of the modified model are still in place, with the survival point weakening the relation between problem-solving response and organizational change. The kinked-curve relation is produced by the survival point and self-enhancement, and does not need to be shown separately. A self-enhancing response means less change than one would observe if a problem-solving response prevailed but it is not the same as rigidity. A manager who seeks to self-enhance might be making changes that are symbolic and also changes that are seen as supplementary to past actions rather than as rejections. How to identify the subtle ways in which self-enhancement might become manifest will require additional efforts to those we have deployed here. Longitudinal analyses of firms are often best suited for capturing the binary relation of change versus no change, missing the element of interpretation that skillful managers exploit when showing that they are changing, but in ways that demonstrate that the past actions were correct even though performance was below the aspiration level. We will discuss the research opportunities that the extended model opens up in Section 7, but first we use Section 6 to give a practical example of organizational behavior that is best understood using its concepts and relations.

7.1 Research on Mechanisms

Translating our theorized mechanisms into research is straightforward and offers the opportunity to make quick progress. A key step here is to examine whether ambiguity enables self-enhancement. This matters because both inside and outside the organization, formalization of the assessment criteria removes ambiguity and hence forces the decision-maker to respond to performance on a specific goal, whereas goal ambiguity allows detachment from a goal and possibly also reduces personal involvement. There is already research showing that ambiguous goals, aspiration levels, or feedback allow self-enhancement and hence inaction in the face of performance below aspiration levels on a specific goal (Reference Audia, Brion and GreveAudia et al., 2015; Reference Joseph and GabaJoseph and Gaba, 2015; Reference Smith and ChaeSmith and Chae, 2017). An obvious and important extension of this research is examination of organizations with complex goals such as hybrid organizations that pursue commercial and social goals (Reference Battilana and DoradoBattilana and Dorado, 2010; Reference Boone and ÖzcanBoone and Özcan, 2016; Reference Smith and BesharovSmith and Besharov, 2017). Much of the literature on hybrid organizations focuses on internal arrangements that enable organizations to pursue incompatible goals seen as having equal importance but is less concerned with whether and how decision-makers make changes in response to performance on these multiple goals. Interestingly, this literature highlights flexibility of interpretation and responses as a key feature of hybrid organizations. Flexibility of interpretation and responses is also a key feature in the extended model. There is, however, a difference in emphasis on the source of this flexibility. While we view flexibility as stemming from individual decision-makers’ propensity to switch from a problem-solving orientation to a self-enhancing orientation, research on hybrid organizations gives greater importance to flexibility motivated by a commitment to preserve the hybrid identity of the organization. Analyses of how hybrid organizations respond to performance on multiple goals may allow additional testing of propositions advanced in this rich and growing literature while simultaneously exploring points of convergence and tension with the extended model of performance feedback.

Studies of hybrid organizations should only be considered a first step because multiple goals are central to many organizations and positions in organizations. An example of significant interest to scholars in organization theory is organizations with profitability and safety goals, such as organizations operating dangerous technologies (Reference Baum and DahlinBaum and Dahlin, 2007; Reference Gaba and GreveGaba and Greve, 2019). The more general case is organizations that have profitability goals but also have brands built on costly activities such as maintenance of high quality. Consumer-facing organizations such as car makers (Reference Haunschild and RheeHaunschild and Rhee, 2004; Reference Hu and BettisHu and Bettis, 2018), restaurants (Durand, Rao, and Monin, 2007), hotels (Reference Wang, Wezel and ForguesWang, Wezel, and Forgues, 2016), food and beverage makers (Reference Hsu, Roberts and SwaminathanHsu, Roberts, and Swaminathan, 2012; Reference MalterMalter, 2014), and other makers of branded goods are in a continuous tug-of-war between the profitability goal and the costly but important quality goal. We should know more about the conditions that influence how decision-makers in these organizations respond to performance on multiple goals.

Research on ambiguous goals is often conducted at the organizational level of analysis because the goals are more clearly specified and behaviors are more observable at this level. This has led to a neglect of a larger phenomenon that concerns the behavior of more organizational participants: At any organizational level we find individuals filling roles that involve a mixture of goals. These goals may be specified clearly or not and may be incentivized or not, but they are usually well known to the individual. To the extent that they are formalized, the extended model predicts that a problem-solving orientation will follow. What is less frequently known is how decision-maker prioritizes multiple goals when the priority order is not formalized, leading to significant goal ambiguity.

Research on how individual organizational members react to performance feedback when there is goal ambiguity is especially urgent because the effects on employees can include decisions with dramatic consequences. We are writing this in the context of the COVID-19 pandemic after having read press reports that nearly all cardiopulmonary resuscitation (CPR) procedures to save advanced-stage patients have been unsuccessful and have probably been harmful to patients neglected as a result of the concentration of medical staff around each CPR event. The press also reports that very few advanced-age and prior-condition patients connected to a ventilator have survived, again to the detriment of others because ventilators have been scarce. Medical staff at all levels understand that saving the patient on their shift and in their room is important, but so is reducing the overall impact of an epidemic. How do these multiple goals with ambiguous trade-offs guide their decisions?

Fortunately, most organizational participants face less dramatic choices, but they still handle multiple goals with some ambiguity. Future research could draw on the extended model to tackle questions like the following: What is the effect of (1) the sheer number of goals the decision-maker is aware of, (2) the proximity of each goal to the decision-maker (e.g., ROA is an organizationally remote goal for most decision-makers), (3) the ambiguity in operationalizing and tracking performance within each goal, and (4) the ambiguity in preference ordering among the goals at any given time. These questions hark back to an old and neglected line of learning research on how choices are influenced by how goals are specified and measured (Reference MarchMarch, 1978). Economists have been attracted to the question of how incentives coupled to any given goal affect decisions (e.g., Reference GibbonsGibbons, 1998), but progress on this single-goal question is limited unless these four fundamental questions on how goals interrelate are answered well. It is very rare that a decision-maker faces only one goal.

The degree of centralization of decisions that impact subunits is related to the abovementioned questions and introduces additional opportunities for future research. Our extended model proposes that, when centralization is greater and thus top managers participate in subunit decisions, a problem-solving orientation is more likely because top managers by virtue of their lesser involvement in subunit activities perceive low subunit performance as less threatening than subunit managers do. This draws on the idea that personal involvement is a key condition prompting self-enhancement. Centralization, however, may introduce other complications. First, centralization of decisions can detach the subunit manager from those decisions and lead to protective self-enhancing cognitions and behaviors – most obviously, by allocating responsibility for low performance to centralized decisions and claiming high performance along other goal dimensions related to actions still under the subunit manager’s control. Second, centralization through incentive schemes prescribes actions indirectly, but still so clearly that the subunit manager may allocate responsibility for low performance to the incentive system rather than to own decisions, and so engage in self-enhancement. These are just two examples of the broader idea that self-enhancing responses can take a variety of forms even under conditions that our model expects to be conducive to a problem-solving orientation. For this reason, we view an examination of the different ways in which individuals in organizations self-enhance as an important research direction.

The link between personal involvement and the motivational orientation decision-makers adopt in response to performance on multiple goals raises important questions regarding firm governance mechanisms. An impression given by work on this topic (Reference Westphal and BednarWestphal and Bednar, 2005; Reference Tuggle, Schnatterly and JohnsonTuggle, Schnatterly, and Johnson, 2010a; Reference Tuggle, Sirmon, Reutzel and BiermanTuggle et al., 2010b) is that an active board of directors compensates for the deficiencies of CEOs that are passive in the face of low organizational performance. Nonetheless, this interpretation overlooks a possible scenario raised by our theory: Active governance by the board may elevate personal involvement to the point where board members may feel as threatened by low performance as the top managers they are expected to control. If boards are more likely to engage in self-enhancement when things go wrong, who will then solve the problems evidenced by low performance? The dilemma is that a board should be sufficiently involved to deter management from self-enhancement behaviors, but not so involved that it also engages in self-enhancement. Governance research has not taken this scenario into account, and accordingly its recommendations may require additional scrutiny. We are noting this not to claim that CEOs should not be governed, but rather to make the point that under conditions of elevated involvement by the board such governance has a cost: biased assessments of performance arising from self-enhancement.

Formalization of performance assessment includes assignment of responsibility to the subunit manager both for actions taken and for performance on one or multiple goals, and this should lead to a stronger problem-solving orientation. Again, this is the theoretical expectation, and we currently do not have sufficient evidence to claim that it holds. We do have clear opportunities for research to examine the effects of formalization on firm and subunit behaviors, thanks to data that are available or easy to collect. For example, governance reform initiatives have led to boards adopting formal procedures for evaluating directors and board decision processes (Reference Rowley, Shipilov and GreveRowley, Shipilov, and Greve, 2017; Reference Shipilov, Greve and RowleyShipilov et al., 2019), and these are likely to be reflected in greater transparency to investors and closer examination of CEO decisions. Many environmental groups create formalized assessments and rankings of organizations, such as the J.D. Powers quality assessments of automobiles, the ISO 9000 process quality assessment (Reference Terlaak and KingTerlaak and King, 2006), and the KLD rankings of corporate social responsibility (Reference Chatterji, Levine and ToffelChatterji, Levine, and Toffel, 2009). Other formalized assessment processes are generated through automatic customer feedback, such as those provided by online platforms (Reference Wang, Wezel and ForguesWang et al., 2016; Reference Greve and SongGreve and Song, 2017).

These assessments and rankings create three kinds of variation. The first is that many of them cover only some firms, either because they are voluntary (e.g., ISO 9000) or because they are done selectively (e.g., KLD). The second is that some assessments do not rank firms, whereas others do and hence are both a formalized assessment and potentially a goal (high rank) in themselves. The third is that numerical ratings or rankings can place firms high or low, and this can affect self-enhancement versus problem-solving just as powerfully as the formalization of a ranking (Reference Rowley, Shipilov and GreveRowley et al., 2017). Given the increased prevalence of ratings systems and their proven effects across industries (Reference Elsbach and KramerElsbach and Kramer, 1996; Reference SauderSauder, 2006; Reference Sauder and EspelandSauder and Espeland, 2009), there is room for much more research on how they affect the self-enhancement versus problem-solving orientation in firms with performance below aspiration levels.

At the subunit level, researchers face the problem of many organizations not being sufficiently open about their formalized performance assessment to allow easy research access. Some of these issues can be resolved by modifying experiments on self-enhancing reactions to performance feedback (e.g., Reference Audia and BrionAudia and Brion, 2007; Reference Jordan and AudiaJordan and Audia, 2012; Reference Audia, Brion and GreveAudia et al., 2015) to also include conditions with different forms and levels of formalized assessment. However, organizational research is also rich with empirical work made possible by firms giving access to their internal records (see Reference BidwellBidwell, 2011; Reference KleinbaumKleinbaum, 2012; Reference CastillaCastilla, 2015), and such research should be pursued.

Research on the effects of multiple institutional logics is closely related to the work on goal ambiguity covered earlier because the mechanism is similar, with multiple institutional logics giving rise to goal ambiguity and often also to organizational forms that have hybrid features. These two research streams can learn from each other and will often unite in a single investigation. The research on multiple logics has a domain that is larger because even organizations that are designed to have a single goal can become embedded in environments with multiple logics, through no initiative of their own, and accordingly face the possibility of responding to prescriptions and goals that other actors hold for them (Reference Greve and TehGreve and Teh, 2018).

We already have some knowledge on organizational responses to multiple external pressures through work done in related areas of research. Institutional logics research (Reference Greenwood, Raynard, Kodeih, Micelotta and LounsburyGreenwood et al., 2011; Reference Thornton, Ocasio and LounsburyThornton, Ocasio, and Lounsbury, 2012) and social movement research (Reference Davis, McAdam, Scott and ZaldDavis et al., 2005; Reference Briscoe and GuptaBriscoe and Gupta, 2016) are rich sources of ideas on how institutional environments seek to impose themselves on organizations. So far, they have not progressed in examining responses to performance feedback, however, because most work in these traditions sees these institutional pressures as main effects rather than as modifiers to how organizations respond to their most valued goals (such as profitability). The extended model of performance feedback complements that work because it outlines a more agentic view of how decision-makers respond not simply to performance feedback but also to cultural prescriptions of what is appropriate. Specifically, decision-makers with a self-enhancement orientation may “use” what macro theorists generally conceive as external constraints to form favorable interpretations of ambiguous information. Indeed, ironically, our extended model implies that attempts to make organizations do more for society through the increasing influence of multiple logics can provide decision-makers with opportunities to do less. This point of intersection between the theory of performance feedback and macro research on logics is so important that work on how goal proliferation affects responses to performance feedback is urgently needed.

The theoretical expectation is that managers facing complex institutional environments with multiple goals in addition to the usual firm goals have significant leeway in playing these goals against each other. For example, research has already found low-profitability firms deciding not to engage in governance reform because profitability improvement is urgently needed or perhaps because the governance reform is not a valued goal (Reference Rowley, Shipilov and GreveRowley et al., 2017). Facing the internal profitability goal and the external governance reform goal introduced sufficient goal ambiguity to let firms avoid problem-solving. Conversely, observing other organizations failing on environmental goals can trigger problemistic search in organizations that are not experiencing performance shortfall (e.g., Reference DesaiDesai, 2014), suggesting that it is especially easy to trigger problemistic search when the organizational decision-makers have no reason to engage in self-enhancement.

Indeed, goal ambiguity can also originate from a mixture of internal goals that are difficult to reconcile or prioritize. Research on such mixtures of goals has just barely started, but has indicated that organizations are less responsive, suggesting that some degree of self-enhancement occurs (Reference Hu and BettisHu and Bettis, 2018). Irreconcilable internal goals currently present an unsolved problem for the theory and evidence. If the theoretical reasoning and modest evidence from internal and external goal conflicts translates into conflicts among internal goals, the potential for decision-makers to exploit the resulting ambiguity to avoid problemistic search is high.

Research on the effects of power is important because power is the mechanism that can enable resistance against a problem-solving approach when the performance is below aspiration levels and either low centralization or high formalization prevents self-enhancing responses. To advance such research, researchers need to address the question of how to measure the power of organizational decision-makers. Fortunately progress has been made in measuring organizational power, much of it based on resource dependence theory (Reference Pfeffer and SalancikPfeffer and Salancik, 1978; Reference Wry, Cobb and AldrichWry, Cobb, and Aldrich, 2013), and these tools can be used along with advances from other research streams. Later work has added important details such as CEO cooptation of the board through selective recruitment of loyal board members (Reference Westphal and ZajacWestphal and Zajac, 1995). This work has shown that the type of boards with great CEO control – because directors are demographically similar to the CEO or have friendship ties to the CEO – result in less changes in response to low performance, consistent with CEO self-enhancement (Reference Westphal and BednarWestphal and Bednar, 2005). Indeed, the work on CEO power relative to the board is quite rich in discovered mechanisms, including such subtle but effective mechanisms as control over the board agenda (Reference Tuggle, Sirmon, Reutzel and BiermanTuggle et al., 2010b; Reference Joseph, Ocasio and McDonnellJoseph, Ocasio, and McDonnell, 2014), with greater agenda control allowing the CEO to respond less to performance below the aspiration level, as a self-enhancement orientation would predict.

7.2 Theoretical Comparisons

The extended model of performance feedback differs from analyses that take on a more stringent cognitive characterization of action (e.g., Reference OcasioOcasio, 1997). Theories that emphasize the cognitive component generally hold that beliefs guide action and that belief updating stems from unbiased assessment of evidence. Our extended model, by acknowledging the influence of personal involvement on motives, implies that, as Reference 58KundaKunda (1987: 637) puts it, “the cognitive apparatus is harnessed in the service of motivational ends.” Whereas individuals not personally involved in an activity display unbiased assessments of negative information and let their preexisting beliefs guide action, those who are personally involved alter their beliefs to reduce self-threat and let the changed beliefs guide action. Focusing on responsiveness to multiple goals, beliefs about which goals are most critical may change as a result of salient threats to the self-image. The contrast between these two approaches hinges on whether motivation can override pure cognitive effects, and experimental work manipulating these factors, similar to Reference Smith and ChaeSmith and Chae (2017) and Reference Audia, Brion and GreveAudia et al. (2015), will help test the contingencies outlined in the extended model and identify boundary conditions.

The extended model also differs from analyses that view identity as a constraint on action. In those views, identity is a permanent filter on what actors see as relevant to define how to act. These views are related to the old insight that experience colors interpretation of events but take the further step of assuming a personally felt identity associated with experience, increasing the rigidity. We do not dispute that such effects are found. Our model implies, however, that actors may be willing to put their identity aside to give attention to features of the situation that help them form favorable evaluations of their self-image or gain rewards in the pursuit of goals. Motivation prevails over identity or at least prompts decision-makers to think of identity in more flexible terms. Again, personal involvement plays a critical role since it is the feature of our model that ties the tendency to protect one’s self-image to self-enhancing assessments. When personal involvement is absent or weak, identity has a stronger impact on how the situation is assessed. Thus, in our model the identity effect is not “always on”; it is triggered by lack of personal involvement. For example, a CEO with a background in finance, who generally gives greater importance to financial metrics, may not let her finance identity inhibit the self-enhancing propensity to elevate in importance marketing goals that are producing better outcomes than financial goals. Identity-based theorizing assumes temporal consistency in preferences, whereas our extended model recognizes the likelihood of temporal inconsistency under conditions of high personal involvement. There is much research on how strong organizational identities shape behaviors, but these implications of our extended model have not been investigated.

At the individual, organizational, and environmental level, our extended model posits different mechanisms that alter the reaction to performance feedback, as one should expect given the difference of these levels of analysis. The alert reader will have noticed a common factor though: power. Taking an apparent problem seriously, searching for a solution, and enacting a solution that has been found are all influenced by power in our extended model. As our discussion showed, the expected effect is not necessarily that more power leads to more action. This is an important theoretical implication that needs empirical testing. More broadly, it fits a movement toward greater considerations of how power affects organizations (Reference Wry, Cobb and AldrichWry et al., 2013). It still differs from the usual resource dependence predictions because, in the extended model, power counteracts factors that force decision-makers into problem-solving. Hence, power is unimportant if performance feedback indicates that there is no problem to solve. Also, it is less important if decision-maker self-enhancement is already enabled, as one would expect from, for example, a decision-maker handling ambiguous goal dimensions and lacking formalized performance assessment. The effects of power on decision-maker orientation are not needed to resist urgent calls for problem-solving because ambiguity and lack of formalization means that there are no such calls to begin with.

7.3 Conclusion

The core model of performance feedback is a fundamental model of organizational behavior. It proved to be broadly applicable, but also to be overly parsimonious, leading to the modified model. The modified model is also fundamental and broadly applicable, and for a wide range of phenomena it predicts and describes outcomes very well. Still, the need to explicitly incorporate multiple goals has led us to propose the extended model we describe here. The addition of tensions between problemistic search and self-enhancement has not made the model any less fundamental, and we believe it has extended the range of phenomena it predicts and explains well. As a result, future research has two main tasks. The first is to test whether the model is correct as specified here. This section lists our suggestions on how to conduct such tests, but the model is so flexible and fundamental that many more are possible. The second is to explore its boundaries. What organizational behaviors can be understood using this model, as opposed to the simpler core or modified model? How might incorporation of the extended model improve other theories that are loosely built on the behavioral theory of the firm?

In the long run, the second of these questions might be the most important. After all, the behavioral theory of the firm in general, and especially problemistic search, is incorporated as an assumption in many organization theory models. The theory used to derive mimetic isomorphism rested on the behavioral theory of the firm (Reference DiMaggio and PowellDiMaggio and Powell, 1983). Adaptive views of the organizational behaviors such as resource dependence rely on problem-solving reasoning (Reference Pfeffer and SalancikPfeffer and Salancik, 1978). In the heyday of organizational ecology, problemistic search was often used as a micro-theory of organizational action (Reference Barnett, Greve and ParkBarnett, Greve, and Park, 1994). Some papers on network and alliance theory assume that performance feedback guides establishment of interorganizational ties (Reference Baum, Rowley, Shipilov and ChuangBaum et al., 2005). What these treatments have in common is reliance on the core model rather than the more elaborate models that have been developed since. How will the theories change if they are integrated with performance feedback as currently modeled? Taking that step will allow theoretical and empirical advances with potential so high that we cannot assess it now.